Hey Sam
I understand your concern. How about a compromise whereby I remove the
_SET flags from the system interface flags but retain something similar
for the wire protocol?
The reason I would like to have these flags is,
there are 3 cases
a) Dont set atime/mtime
b) Set atime/mtime to whatever time the server chooses
c) Forcibly set atime/mtime to the value provided.
If we have only 1 flag, how do we differentiate between these 3 cases?

The server would not know when (a) happens. Only client does.
Let me know if that is ok?
Thanks,
Murali

On Mon, 31 Jul 2006, Sam Lang wrote:

>
> Hi Murali,
>
> Patch looks good overall.  Would it be possible to reuse the _ATIME,
> _CTIME, and _MTIME options instead of defining _ATIME_SET,
> _CTIME_SET, and _MTIME_SET to explicitly set the times to the values
> passed in?
>
> In other words, are there any cases where we call
> PVFS_object_attr_overwrite_setable where instead we could
> individually set the appropriate *time based on the context (which
> server operation we're doing)?
>
> I'm looking to avoid adding more attr flags to the system interfaces
> and changing the setattr protocol.
>
> -sam
>
> On Jul 30, 2006, at 11:49 AM, Murali Vilayannur wrote:
>
> > <mtime2.patch.gz>
>
>
_______________________________________________
Pvfs2-developers mailing list
[email protected]
http://www.beowulf-underground.org/mailman/listinfo/pvfs2-developers

Reply via email to